Search

Objectives Damage accrual in SSc can be tracked using the Scleroderma Clinical Trials Consortium Damage Index (DI). Our goal was to develop a prediction model for damage accrual in SSc patients with early disease. Methods Using patients with <2 years disease duration from Canada and Australia as a derivation cohort, and from the Netherlands as a validation cohort, we used group-based trajectory modelling (GBTM) to determine 'good' and 'bad' latent damage trajectories. We developed a prediction model from this analysis and applied it to patients from derivation and validation cohorts. We plotted the actual DI trajectories of the patients predicted to be in 'good' or 'bad' groups. Results We found that the actual trajectories of damage accumulation for lcSSc and dcSSc were very different, so we studied each subset separately. GBTM found two distinct trajectories in lcSSc and three in dcSSc. We collapsed the two worse trajectories in the dcSSc into one group and developed a prediction model for inclusion in either 'good' or 'bad' trajectories. The performance of models using only baseline DI and sex was excellent with ROC AUC of 0.9313 for lcSSc and 0.9027 for dcSSc. Using this model, we determined whether patients would fall into 'good' or 'bad' trajectory groups and then plotted their actual trajectories which showed clear differences between the predicted 'good' and 'bad' cases in both derivation and validation cohorts. Conclusions A simple model using only cutaneous subset, baseline DI and sex can predict damage accumulation in early SSc. [ABSTRACT FROM AUTHOR]

Background and aims: The experience of alcohol use among adolescents is complex, with international differences in age of purchase and individual differences in consumption and consequences. This latter underlines the importance of prediction modeling of adolescent alcohol use. The current study (a) compared the performance of seven machine‐learning algorithms to predict different levels of alcohol use in mid‐adolescence and (b) used a cross‐cultural cross‐study scheme in the training‐validation‐test process to display the predictive power of the best performing machine‐learning algorithm. Design: A comparison of seven machine‐learning algorithms: logistic regression, support vector machines, random forest, neural network, lasso regression, ridge regression and elastic‐net. Setting: Canada and Australia. Participants: The Canadian sample is part of a 4‐year follow‐up (2012–16) of the Co‐Venture cohort (n = 3826, baseline age 12.8 ± 0.4, 49.2% girls). The Australian sample is part of a 3‐year follow‐up (2012–15) of the Climate Schools and Preventure (CAP) cohort (n = 2190, baseline age 13.3 ± 0.3, 43.7% girls). Measurements: The algorithms used several prediction indices, such as F1 prediction score, accuracy, precision, recall, negative predictive value and area under the curve (AUC). Findings: Based on prediction indices, the elastic‐net machine‐learning algorithm showed the best predictive performance in both Canadian (AUC = 0.869 ± 0.066) and Australian (AUC = 0.855 ± 0.072) samples. Domain contribution analysis showed that the highest prediction accuracy indices yielded from models with only psychopathology (AUC = 0.816 ± 0.044/0.790 ± 0.071 in Canada/Australia) and only personality clusters (AUC = 0.776 ± 0.063/0.796 ± 0.066 in Canada/Australia). Similarly, regardless of the level of alcohol use, in both samples, externalizing psychopathologies, alcohol use at baseline and the sensation‐seeking personality profile contributed to the prediction. Conclusions: Computerized screening software shows promise in predicting the risk of alcohol use among adolescents. [ABSTRACT FROM AUTHOR]